The advantage of U2F is that it uses an asymmetric scheme where a keypair is generated for each authentication endpoint, so the secrets never leave the hardware token in the first place. Compared to RSA SecurID, you would need to have a really severe threat scenario such that you can't even trust RSA with your keys. Of course, U2F is also a much more complicated protocol than say TOTP, and has not existed long enough to stand the cryptographic test of time.
To answer your second question, I'm not sure I understand what you mean. TOTP (Google Authenticator), RSA SecurID, and U2F are all different types of 2FA. They could all be combined with Kerberos if that is what you want.
Such as the one where RSA was compromised and secrets were leaked, sadly.
Wouldn't a more elegant solution be to add Kerberos authentication combined with 2FA? You can then pay the price for access up front but someone phishing your password gets nothing since presumably they cannot answer the second factor challenge.